A Keyword-Based Method for Measuring Sentence Similarity
A Keyword-Based Method for Measuring Sentence Similarity
复制标题
一种基于关键词的句子相似度测量方法
DOI：
10.1145/3091478.3098878
复制
发表时间：
2017
期刊：
影响因子：
--
通讯作者：
Jinxing Cheng
中科院分区：
文献类型：
--
作者：
Yuanjun Bi;Kai Deng;Jinxing Cheng
提出了一种基于关键词的句子相似度计算方法。首先，它从句子中提取关键字。然后，该方法计算关键字的排名分数。最后利用Jaccard相似度系数将排序结果应用到句子相似度计算中。在一个真实的Chatterbot系统数据集上的实验表明，该方法将句子相似度方法的相关度提高了30%。
In this paper, a sentence similarity computing approach based on keywords is presented. First, it extracts the keywords from a sentence. Then the approach computes ranking scores for the keywords. Finally it applies these ranking scores into the sentence similarity computation using the Jaccard similarity coefficient. Experiments on a real word chatterbot system dataset demonstrate that this proposed approach significantly improves the relevance of sentence similarity method up to 30%.